THE BLAST FURNACE. 51 that with pure lime and a silica content of 39 to 40 per cent, the cinder "slacked," but with dolomite the silica could be reduced to 35 per cent, and the furnace worked better. He refers to various investigators who claim that a high content of magnesia gives rise to the production of spinel, an infusible compound of alumina., lime and magnesia,, and argues that the formation of this mineral depends upon the presence of a large proportion of alumina, as well as magnesia, so that no harm will result from 20 per cent, of magnesia in the slag if the alumina is under 10 per cent. In regard to alumina, it is stated by Elbers* that if the percentage of silica be low it acts as an acid, and hence increases the fluidity of the slag, but if high it acts as a base, and thus lowers the fusing point. Phillipst says that in every-day practice and with slags of 33 and 36 per cent, silica, the alumina is considered as silica. Many of the accurate limitations set by special investigators after a limited series of experiments are erroneous. Thus I have the slag records of a furnace for four months, where the cinder was fairly constant and averaged as follows, in per cent.: Si02, 35 A1203, 14.5 CaO, 28 MgO, 22 This upsets any theory that high alumina and high magnesia are incompatible. In the same way, experiments made at Steelton show that alumina can be carried above 35 per cent, with perfect elimination of the sulphur and good working of the furnace, and it appears to replace, to some extent, both silica and lime, and may therefore be regarded as merely passively diluting the cinder. This will be evident from the following series of slags, arranged in order of increasing alumina. Bach column is the average of several casts from a furnace operating for over a week on an aluminous burden. Si02 .................... 34 34 29 25 A1203.................... 10 16 27 33 CaO plus MgO............ 54 45 40 39 The general range of blast-furnace slags is illustrated in Table II-A. *Berg- und Huttenmannische Zeitung, Vol. XLVII, p. 253. t Ala. Geol. Survey, 1898, p. 45.